Multimodal Content Generation on Human Evaluation N=20 (test)
19Win CountCogGen
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| CogGenDimension=Content Depth, Comparison Baseline=MMDR2026.04 | 19 | — | — | — | — | 0 | 1 | 95 | |
| CogGenDimension=Overall Quality, Comparison Baseline=MMDR2026.04 | 18 | — | — | — | — | 1 | 1 | 90 | |
| CogGenComparison Baseline=Gemini Deep Research, Evaluation Dimension=Visual-Text Alignment, Sample Size=N=202026.04 | 16 | — | 80 | — | — | 0 | 4 | 80 | |
| CogGenComparison Baseline=Gemini Deep Research, Evaluation Dimension=Multimodal Synergy, Sample Size=N=202026.04 | 16 | — | — | 80 | — | 1 | 3 | 80 | |
| CogGenDimension=Visual-Text Alignment, Comparison Baseline=MMDR2026.04 | 16 | — | — | — | — | 2 | 2 | 80 | |
| CogGenDimension=Multimodal Synergy, Comparison Baseline=MMDR2026.04 | 16 | — | — | — | — | 2 | 2 | 80 | |
| CogGenComparison Baseline=Gemini Deep Research, Evaluation Dimension=Overall Quality, Sample Size=N=202026.04 | 15 | 75 | — | — | — | 1 | 4 | 75 | |
| CogGenComparison Baseline=Gemini Deep Research, Evaluation Dimension=Content Depth, Sample Size=N=202026.04 | 10 | — | — | — | 50 | 3 | 7 | 50 |